• Version
• Download 19
• File Size 581.51 KB
• File Count 1
• Create Date 2 May 2025
• Last Updated 2 May 2025

Prediction of At-Risk Students in E-Learning Platforms Using Deep Learning Models

Authors:

1. RUPADEVI¹, GANGAIAHGARI RAMBHAVANI ²

¹Associate Professor, Dept of MCA, Annamacharya Institute of Technology & Sciences, Tirupati, AP, India, Email:rupadevi.aitt@annamacharyagroup.org

²Post Graduate, Dept of MCA, Annamacharya Institute of Technology & Sciences, Tirupati, AP, India, Email: greddybhavani17@gmail.com

Abstract: Virtual learning environments have emerged as a major educational tool in recent years. But because of this shift, schools now have a difficult time identifying and assisting kids who are at risk of dropping out. Increased dropout rates result from the inability to engage with pupils in person, which delays prompt assistance. This work offers a hybrid deep learning model intended to forecast high-risk students in online learning environments to overcome this difficulty. Important academic and behavioural variables, such as prior academic performance, weekly login frequency, average session duration, quiz results, and final exam performance, are used in our model. To increase prediction accuracy, these features are run through a hybrid deep learning framework that makes use of both deep neural networks and conventional machine learning elements. A user-friendly Flask-based web application is used to implement the system, enabling real-time predictions and educator participation. Early intervention and improved student support may result from the suggested model's promising performance in detecting at-risk kids. This study is a useful resource for online educational institutions since it blends sophisticated prediction methods with real-world implementation.

Keywords: Student Risk Prediction, Virtual Learning, Hybrid Deep Learning, Dropout Detection, Educational Analytics.